Cell Stress and Chaperones https://doi.org/10.1007/s12192-018-0897-y

ORIGINAL PAPER

Amyotrophic lateral sclerosis (ALS) and Alzheimer’s disease (AD) are characterised by differential activation of ER stress pathways: focus on UPR target

L. Montibeller1 & J. de Belleroche1

Received: 23 January 2018 /Revised: 6 April 2018 /Accepted: 8 April 2018 # The Author(s) 2018

Abstract The endoplasmic reticulum (ER) plays an important role in maintenance of proteostasis through the unfolded response (UPR), which is strongly activated in most neurodegenerative disorders. UPR signalling pathways mediated by IRE1α and ATF6 play a crucial role in the maintenance of ER homeostasis through the transactivation of an array of transcription factors. When activated, these transcription factors induce the expression of genes involved in and degradation with pro-survival effects. However, the specific contribution of these transcription factors to different neurodegenerative diseases remains poorly defined. Here, we characterised 44 target genes strongly influenced by XBP1 and ATF6 and quantified the expression of a subset of genes in the post-mortem spinal cord from amyotrophic lateral sclerosis (ALS) cases and in the frontal and temporal cortex from frontotemporal lobar degeneration (FTLD) and Alzheimer’s disease (AD) cases and controls. We found that IRE1α- XBP1 and ATF6 pathways were strongly activated both in ALS and AD. In ALS, XBP1 and ATF6 activation was confirmed by a substantial increase in the expression of both known and novel target genes involved particularly in co- activity and ER-associated degradation (ERAD) such as DNAJB9, SEL1L and OS9. In AD cases, a distinct pattern emerged, where targets involved in protein folding were more prominent, such as CANX, PDIA3 and PDIA6. These results reveal that both overlapping and disease-specific patterns of IRE1α-XBP1 and ATF6 target genes are activated in AD and ALS, which may be relevant to the development of new therapeutic strategies.

Keywords UPR . ALS . AD . ERAD . Folding . ER stress

Abbreviations FTLD Frontotemporal lobar degeneration AD Alzheimer’sdisease IRE1 Inositol-requiring enzyme 1 ALS Amyotrophic lateral sclerosis NFE2L2 Nuclear factor, erythroid 2-like 2 ATF4 Activating transcription factor 4 PDI Protein disulphide isomerase ATF6 Activating transcription factor 6 PERK Protein kinase RNA-like endoplasmic BiP Binding immunoglobulin protein reticulum kinase DNAJs DnaJ family PS1 Presenilin-1 ER Endoplasmic reticulum qPCR Quantitative polymerase chain reaction ERAD ER-associated degradation TF Transcription factor ERSE ER stress response elements UPR Unfolded protein response UPRE Unfolded protein response element XBP1 X box-binding protein 1 Electronic supplementary material The online version of this article (https://doi.org/10.1007/s12192-018-0897-y) contains supplementary material, which is available to authorized users. Introduction * J. de Belleroche [email protected] The majority of neurodegenerative diseases such as 1 Neurogenetics Group, Division of Brain Sciences, Faculty of Alzheimer’s disease (AD), amyotrophic lateral sclerosis Medicine, Imperial College London, London, UK (ALS) and frontotemporal lobar degeneration (FTLD) are L. Montibeller, J. de Belleroche characterised by dysfunction in protein homeostasis response element) (Yamamoto et al. 2004; Acosta-Alvear (proteostasis) (Hetz and Saxena 2017). A major role in et al. 2007; Misiewicz et al. 2013). A highly conserved core proteostasis is played by the endoplasmic reticulum (ER), region (ACGT core) which has been found as a single entity which is responsible for the correct folding, modification and within the UPRE sequence is also recognised by XBP1s and quality control of approximately one third of all secretory (Kanemoto et al. 2005; Acosta-Alvear et al. 2007). Notably, and membrane (Hetz and Saxena 2017). When the several studies have demonstrated that ERSE and ERSE-II protein folding demand exceeds the capacity of the cell to binding sites are also recognised by ATF6 (Yamamoto restore homeostasis, misfolded proteins accumulate in the 2004), which may be able to hetero-dimerise with XBP1 ac- ER leading to a condition known as ER stress. Markers of tivating different sets of genes such as DNAJC3 and HYOU1 ER stress are evident in many neurodegenerative diseases in- (Newman 2003; Yamamoto et al. 2007; Shoulders et al. 2013). dicating the involvement of common homeostatic mecha- Thus, due to this close interaction, the specific contribution of nisms such as the unfolded protein response (UPR) IRE1α-XBP1 and ATF6 pathways in normal and pathological (Kampinga and Bergink 2016). condition is still not well understood. The UPR is a system composed of three intimately con- Recent studies have demonstrated an involvement of the nected signalling pathways, which are mediated by three ER IRE1α-XBP1 and ATF6 pathways in human neurodegenera- membrane-resident sensors: the protein kinase RNA (PKR)- tive diseases. Phosphorylated IRE1α (p-IRE1α) and XBP1s like ER kinase (PERK), the transmembrane basic leucine zip- have been found to be increased in different human brain per activating transcription factor-6 (ATF6) and the evolution- regions of patients affected by AD (Hoozemans et al. 2009; ary conserved kinase/endoribonuclease inositol-requiring pro- Lee et al. 2010). The activation of the IRE1α-XBP1 pathway tein-1 (IRE1α). Within the UPR, IRE1α and ATF6 pathways correlates with the presence of abnormally phosphorylated tau play a pro-survival role mediated by the induction of ER chap- in AD neurons (Hoozemans et al. 2009). Furthermore, the erones and proteins involved in the removal of misfolded pro- molecular activation of this pathway in animal models of teins (Pincus et al. 2010; Shoulders et al. 2013). The regula- AD is involved in the prevention of Aβ toxicity through the tion of these homeostatic genes is mediated by a set of related regulation of specific targets such as HRD1 (Casas-Tinto basic leucine zipper (bZIP) transcription factors which are et al. 2011;EndresandReinhardt2013;Songetal.2015). The activated by each UPR branch through unique signal trans- IRE1α-XBP1 pathway has also been found to be activated in duction mechanisms (Travers et al. 2000). IRE1α is activated the human spinal cord of sporadic cases and in mouse models by autophosphorylation leading to the processing of X box- of ALS (Atkin et al. 2008). This activation was associated binding protein-1 mRNA, to yield (XBP1s), which is a potent with elevated levels of ER chaperones such as PDIA1 and transcription factor (TF) (Acosta-Alvear et al. 2007;Shoulders ERp57 (Ilieva et al. 2007; Hetz et al. 2009). Interestingly, et al. 2013). In addition to XBP1, activated IRE1α is able to the activation of this pathway represents one of the earliest mediate and process the cleavage of cytoprotective mRNAs pathological events in motor neurons, which occurs before through a process called IRE1α-dependent decay (RIDD) the onset of symptoms in mouse models of ALS (Atkin (Maureletal.2014; Tam et al. 2014; Hetz and Saxena et al. 2008). In addition, an increase in XBP1 splicing and its 2017). The ATF6 full-length protein, instead, is activated after translocation to the nucleus have been demonstrated to occur translocation to the Golgi apparatus where it is processed and in motor neuronal cell lines expressing ALS-associated muta- released as a soluble cytoplasmic fragment that is able to enter tions (Prell et al. 2012). This evidence suggests that the into the nucleus and activate the transcription of a specific IRE1α-XBP1 pathway is activated in AD and ALS by in- subset of genes (Ye et al. 2000). There are several lines of creasing the expression of a set of genes that remove aberrant evidence showing that IRE1α-XBP1 and ATF6 pathways proteins and restore protein homeostasis (Casas-Tinto et al. are intimately inter-connected. The expression of XBP1 is 2011). In contrast, the involvement of the ATF6 pathway in partially regulated by ATF6 (Yoshida et al. 2001; Dunham neurodegenerative disorders is poorly described and remains et al. 2012). Moreover, IRE1α-XBP1 and ATF6 pathways largely unknown (Xiang et al. 2017). In AD, a PS1 mutation regulate the transcriptional induction of a similar set of genes, inhibits the activation of ATF6 whereas protein levels of acti- which are characterised by the presence of specific cis-acting vated ATF6 increased in the spinal cord of sporadic ALS elements in their promoter regions (Acosta-Alvear et al. 2007; (Katayama et al. 2001; Atkin et al. 2008). In addition, both Yamamoto et al. 2007; Pincus et al. 2010; Shoulders et al. IRE1α-XBP1 and ATF6 transcriptional activity were found to 2013). These cis-acting elements are recognised by XBP1 be significantly attenuated in cell line models carrying an and can be divided into four main binding sites: ERSE (ER ALS-linkedmutationinVAPB(Chenetal.2010;Nardo stress response elements, the consensus sequence of which is et al. 2011). Thus, although this evidence supports the in- CCAAT-N9-CCACG) (Yoshida et al. 2001), ERSE-II (ER volvement of IRE1α-XBP1 and ATF6 pathways in AD and stress response element II, ATTGG-N1-CCACG), ERSE-26 ALS, their specific role in these diseases remains poorly (CCAAT-N26-CCACG) and UPRE (unfolded protein defined. Amyotrophic lateral sclerosis (ALS) and Alzheimer’s disease (AD) are characterised by differential...

Here, we compared the IRE1α-XBP1 and ATF6 pathways the Brains for Dementia Research Network. All tissue was ob- in human post-mortem tissues of CNS in different neurode- tained from voluntary donors in compliance with the Mental generative disorders. We found a differential activation of Capacity Act (2005), and the Brain Bank has been approved IRE1α-XBP1 and ATF6 target gene sets in ALS compared by the National Research Ethics Service. All clinical diagnoses to AD cases including genes that had never been described were confirmed neuropathologically at post-mortem. Dissected before in these disorders. This study extends our understand- brain tissue was snap frozen, then stored at − 80 °C until further ing of the cellular mechanisms of ER stress involved in neu- use. More details can be found in SI Materials and Methods. rodegenerative disorders highlighting similarities and differ- ences between IRE1α-XBP1 and ATF6 in AD and ALS, Selection of gene targets which can be used to identify therapeutic targets and to devel- op disease-specific targets/biomarkers. We considered XBP1 as the transcription factor for the IRE1α pathway(Yoshidaetal.2001), ATF6 for the ATF6 pathway (Haze et al. 1999) and ATF4/NFE2L2 for the PERK pathway Materials and methods (Cullinan et al. 2003; Fels and Koumenis 2006). The gene names were extracted from three different databases Reagents, oligonucleotide primers and general methods for (JASPAR, Reactome and TRUSST) selecting Homo sapiens bioinformatic analyses are described in SI Materials and genome 19 (hg19) as the background model. More details can Methods. be found in SI Materials and methods.

Subjects Cis-acting element analysis

Spinal cord samples were available for 49 samples: 32 sporadic For each RefSeq sequence of XBP1 target gene, we investi- ALS(SALS)caseswithamedianageatdeathof68years gated the presence of ER stress response elements (if a gene (range 24–85 years) and median post-mortem delay of 14 h showed multiple RefSeq sequences, the prevalent transcript (range 5–25 h) and 17 control cases with a median age at death was considered). Within those sequences 20 kb upstream of 66 years (range 20–91 years) and median post-mortem delay and 5 kb downstream, we looked for exact matches or one of 11 h (range 3–19 h). Frozen dorsolateral prefrontal cortex base mismatch to the ER stress response element (ERSE) (PFC) and temporal cortex (TC) tissues were obtained from 79 (CCAAT-N9-CCACG), ERSE-II (ATTGG-N-CCACG), subjects: 20 were healthy controls with only ageing-related ERSE-26 (CCAAT-N26-CCACG) and the unfolded protein changes and 20 were classified as Alzheimer’s disease (AD) response element (UPRE) (TGACGTGG/A) sequences. cases and 39 as frontotemporal lobar degeneration cases. Of the More details can be found in SI Materials and Methods. 39 FTLD cases, 19 were familial FTLD cases with a C9orf72 hexanucleotide repeat expansion (FTLD C9+ve), while 20 did mRNA extraction and RNA quality assessment not have a C9orf72 hexanucleotide repeat expansion (FTLD C9-ve). Gender, age at death and post-mortem delay (PMD) mRNA was extracted from frozen tissue samples and stored at of subjects are listed in the additional file: Table S1.More − 80 °C. RNA was extracted from frontal and temporal cortex details can be found in SI Materials and Methods. and spinal cord samples following the Direct-zol RNA mini prep (Zymo Research) protocol. RNA purity and integrity for Human tissue sample preparation all tissue samples was assessed by using multiple well- established methods. For all spinal cord samples (ALS cases This study was approved by the Riverside Research Ethics and controls), we measured tissue pH, as brain pH is de- Committee and was carried out according to their guidelines. creased by prolonged death and hypoxia but remains stable For the spinal cord cases and controls, frozen lumbar tissue post-mortem and varies little across regions. In brief, pH was sampled at levels L3 to L5 was used. Characteristics of this measured in frozen cortical tissue samples from all cases and tissue have been described in detail from the point of view of controls. Median values of pH for ALS cases and controls motor neuron counts, Nissl staining, immunohistochemistry of were 6.3 and 6.4, respectively (Anagnostou et al. 2010). neuronal markers (ChAT, VAPB and DAO) and p62 as a marker Assessment of mRNA purity/quality in the total RNA sample of ubiquitinated protein inclusions typical of ALS (Paul et al. was based on measurement of A260/280 absorbance ratios in 2014). In addition, the glial marker, GFAP, is significantly up- all mRNA extractions and were typically in the range of 1.83 regulated in these ALS cases, indicative of astrocytic prolifera- to 1.94 (mean = 1.89) for spinal cord tissue and 1.77–2.05 tion known to occur in ALS (Anagnostou et al. 2010). Frozen (mean = 1.94) for frontal and temporal cortex tissues, indicat- dorsolateral PFC and TC tissues were obtained from the MRC ing a high level of purity. Assessment of RNA integrity was London Neurodegenerative Diseases Brain Bank, a member of based on the integrity of ribosomal RNAs (rRNAs). The L. Montibeller, J. de Belleroche

Agilent 2100 Bioanalyser in conjunction with the RNA 6000 two-way ANOVA, followed by Tukey’s multiple comparison LabChip® was used for assessment of RNA integrity of total tests for multiple group comparisons. Non-normally distribut- RNA extracts from the spinal cord. Total RNAs with 28S:18S ed data were analysed by the corresponding nonparametric rRNA ratios > 1.0 were considered suitable for qPCR analysis tests (Mann-Whitney U test and Kruskal-Wallis followed by and the remainder were rejected from further study. Further Dunn’s post-test, respectively). All data were checked for the information about RNA integrity was obtained from electro- presence of outliers by performing the GraphPad ROUT test. phoresis of all qPCR amplification products and analysis of Sample size was described in the legend of each figure. Where thermal dissociation profiles to ensure that single uncontami- qPCR analysis was carried out in two batches, the means of nated products were obtained (Fig. S5-6). Procedures for each control group were compared and, if different, a normal- cDNA synthesis and mRNA quantification by qPCR are de- isation factor was applied. A P value of less than 0.05 was scribed in SI Materials and Methods. considered significant.

Quantitative real-time PCR Results qPCR was performed using the Power Up™ SYBr™ Green Master Mix with 6 μM of primers. Stratagene® MX3000p Systematic analysis of the unfolded protein response qPCR system Primer sequences and temperatures utilised for target genes and ER stress response elements of XBP1 real-time PCR analysis are listed in the additional file: and ATF6 in human CNS Table S2. There was no significant correlation between mRNA levels and age at death and post-mortem delay in the The activation of the IRE1α-XBP1 and ATF6 pathways is spinal cord or in the frontal and temporal cortex. Similarly, no triggered in ALS as well as other neurodegenerative diseases significant findings were obtained when each group was (Hetz and Saxena 2017). To characterise these specific re- analysed separately, except for one gene, calnexin, which sponses in detail, we extrapolated the target genes of the main showed a significant decline in SALS with age. More details UPR arms (XBP1, ATF6 and PERK) from three different can be found in SI Materials and Methods. databases (JASPAR, Reactome and TRRUST). We consid- ered XBP1 as the transcription factor for the IRE1α pathway Western blot analysis (Yoshida et al. 2001), ATF6 for the ATF6 pathway (Haze et al. 1999) and ATF4/NFE2L2 for the PERK pathway (Cullinan Sections from frozen tissue blocks were cut (15 μm) on a et al. 2003; Fels and Koumenis 2006). We obtained 158 genes cryostat (Bright Instruments), homogenised and lysed with potentially regulated by ATF4-NFE2L2, 63 by ATF6 and 168 lysis buffer (20 nM Tria-HCl, 137 mM NaCl, 10% glycerol, by XBP1s (additional file: Fig. S1A, Table S3). The target 1% NP40, 2 nM EDTA). Proteins were separated with 10% gene lists were merged in order to identify the common and SDS-PAGE, and blotting was carried out using conditions the specific candidates that are regulated exclusively by each specified for the antibodies. Antibodies used were anti- UPR pathway (Fig. 1a). To select the candidates that are reg- HERPUD1 (Cell Signalling Technology), anti-PDIA6 ulated predominantly by IRE1α-XBP1 and ATF6, several pa- (Abcam), anti-HSPA5/GRP78 (Proteintech), anti-DNAJC10 rameters were used (Fig. 1b). Firstly, we refined the list of (Proteintech) and anti-β actin (Proteintech). More details can candidates considering the already well-established genes reg- be found in SI Materials and Methods. ulated by XBP1 (Acosta-Alvear et al. 2007; Sriburi et al. 2007; Dombroski et al. 2010; Shoulders et al. 2013)and Statistical analysis ATF6 (Arai et al. 2006; Yamamoto et al. 2007;Adachietal. 2008; Shoulders et al. 2013; Howarth et al. 2014) (additional Statistical analyses were performed using GraphPad Prism 5 file: Table S4). Secondly, we combined this information with software (GraphPad, La Jolla, CA, USA). Prism and R were the intra-cellular localisation and tissue expression of the can- used to draw graphs. For bioinformatic analyses, modified didates prioritising the genes that encode for proteins with ER Fisher’s exact P value (EASE) was used to determine the localisation and neuronal expression (additional file: gene-enrichment analysis and Benjamini-Hochberg multiple Table S4). Thirdly, we selected the candidate genes that are test correction to calculate the false discovery rate (FDR). mainly involved in the protein homeostasis process, which Only the corrected values with P < 0.01 were analysed. For were extrapolated from (GO) enrichment anal- the analyses, the D’Agostino-Pearson test ysis performed on the genes regulated by the UPR TFs (addi- was used to test the normality of the data distribution. tional file: Fig. S1B-D, Table S5). Finally, we investigated the Results were expressed as mean ± SEM unless otherwise in- presence of ER stress/unfolded protein response elements dicated. All statistical analyses and calculation of P values (ERSE, ERSE-II, UPRE and ERSE-26) in promoter regions were performed using either two-tailed Student’s t test or of the XBP1 and ATF6 target genes (Fig. 1c, d). Following Amyotrophic lateral sclerosis (ALS) and Alzheimer’s disease (AD) are characterised by differential...

Fig. 1 Systematic analysis of target genes of the UPR transcription stacks of symbols, one stack for each position in the sequence. The factors XBP1, ATF6 and ATF4-NRF2. a Venn diagram showing the overall height of the stack indicates the sequence conservation at that exclusive and overlapping target genes regulated by UPR TFs. The position; the height of symbols within the stack indicates the relative common genes between XBP1 and ATF4-NFE2L2 are 9, between frequency of each amino or nucleic acid at that position. A colour code XBP1 and ATF6 are 14, between ATF6 and ATF4-NFE2L2 are 6. Two is used for each base (blue, cytosine; red, thymidine; yellow, guanosine; genes were shared between all UPR TF. b Workflow for the selection of green, adenosine). d The 44 XBP1 target genes listed in Table S4 were XBP1 and ATF6 target genes. c The 5 ER response elements identified in characterised and gathered based on the ER response elements present in the promoter region of 44 XBP1 and ATF6 target genes were aligned their promoter. The genes inside the grey boxes were found to be using WebLogo (http://weblogo.berkeley.edu/). Each logo comprises regulated by both XBP1 and ATF6 according to the literature these criteria, we selected 44 genes and we highlighted those of ALS (Fig. 2a). Frozen tissues from frontal and temporal that were regulated eitherby XBP1 or ATF6 or by both tran- cortex regions (Lindberg et al. 2012; Möller et al. 2016) scription factors (Fig. 1d, additional file: Table S4). were used to assess the IRE1α-XBP1 pathway activation Successively, a subset of 18 target genes was selected as a in AD and FTLD cases. The FTLD cases, all reference to understand the unique and the cooperative roles neuropathologically characterised by TDP-43 inclusions, of XBP1 and ATF6. We included also a gene (SIL1)whichis were divided in two groups based on the presence (FTLD not regulated by neither ATF6 nor XBP1 although it is in- C9+) or absence (FTLD C9−)ofC9orf72 repeat expan- volved in protein folding processes. Finally, the expression sions. C9orf72 repeat expansions are positively correlated of the these 19 genes was investigated in human brain frozen with the risk of familial and SALS and are recognised as the tissues derived by ALS, FTLD and AD cases relative to con- most common genetic cause of ALS and FTLD (Boylan trol cases (Table 1). 2015). Although XBP1s was found to be upregulated in ALS, no changes in XBP1s expression were detected either The IRE1α-XBP1 pathway is activated in ALS and AD in C9orf72-positive or C9orf72-negative FTLD cases (Fig. but not in FTLD cases 2b). Accordingly, the expression of XBP1 target genes such as HSPA5 and DNAJB9 was found to be comparable be- Initially, we investigated the activation of the IRE1α- tween FTLD and control cases (additional file: Fig. S2A). XBP1 pathway in ALS, AD and FTLD cases. To this However, we found a significant upregulation of XBP1s in end, we analysed the expression of the spliced form of both frontal and temporal cortex regions of AD cases (Fig. XBP1 (XBP1s) in the different brain regions associated 2c). These results demonstrated that the IRE1α-XBP1 path- with each disease. In agreement with the literature (Prell way is activated in ALS and AD cases but not in FTLD et al. 2012), we found a substantial increase in XBP1s cases (whether or not a C9orf72 hexanucleotide repeat ex- expression in the spinal cord derived from sporadic cases pansion was present). L. Montibeller, J. de Belleroche

Table 1 Selection of UPR target genes based on the response elements and transcriptional regulation by XBP1 and ATF6

Name RefSeq Response elements Position to TSS Main UPR TF Secondary UPR TF

CANX NM_001024649.1 ERSE − 109 XBP1 DNAJB9 NM_012328.2 CRE-like − 1955 XBP1 ACGT core − 59 DNAJC10 NM_001271581.1 UPRE − 132 XBP1 ERLEC1 NM_015701.4 ERSE-26 − 255 XBP1 P4HB NM_000918.3 ACGT core − 490 XBP1 SYVN1 NM_032431.2 UPRE − 557 XBP1 ERSE − 521 VCP NM_007126.3 URPE − 452 XBP1 ACGT − 18,167 core ERSE-26 +1342 DNAJB11 NM_016306.5 3× ERSE − 357 XBP1, ATF6 ACGT core − 2238 DNAJC3 NM_006260.4 ERSE − 233 XBP1, ATF6 EDEM1 NM_014674.2 ERSE − 411 XBP1, ATF6 HYOU1 NM_006389.4 UPRE − 242 XBP1, ATF6 PDIA3 NM_005313.4 ERSE − 159 XBP1, ATF6 UPRE − 4056 PDIA6 NM_001282704.1 ERSE − 157 XBP1, ATF6 XBP1 NM_005080.3 ERSE − 78 XBP1, ATF6 NM_014685.3 ERSE-II − 167 ATF6 XBP1 HERP- ERSE − 209 UD1 ACGT core − 563 HSPA5 NM_005347.4 3× ERSE − 68 ATF6 XBP1 ERSE-II − 107 PDIA4 NM_004911.4 ERSE − 12,331 ATF6 XBP1 UPRE +6829 OS9 NM_006812.3 ACGT core − 601 AFT6 XBP1 ERSE +328 SEL1L NM_005065.5 ––AFT6 SIL1 NM_001037633.1 –––

The selected IRE1α-XBP1 target genes were characterised based on different parameters. Two parameters are reported: (i) the presence of ERSE elements in their promoter regions, covering 20 kb upstream and 5 kb downstream of ATG, and (ii) evidence for their expression regulation by XBP1 or ATF6 (references detailed are reported in Table S4). In addition, other parameters were considered for the XBP1 target gene selection such as (iii) function, defined by biological process through gene ontology analysis (see Fig. S1,TableS5); (iv) intra-cellular localisation, obtained by different databases (see BMaterial and methods^); and (v) the association with neurodegenerative diseases, in particular with ALS, AD and FTLD (see Table S4). Among the 18 target genes selected, 7 genes are regulated exclusively by XBP1, 1 exclusively by ATF6 and 10 regulated by both TFs. In addition, one gene (SIL1) was selected because it is not regulated by neither XBP1 nor ATF6 ERSE ER stress response element, UPRE unfolded protein response element

Characterisation of the expression of XBP1 and ATF6 and PDIA4 was increased in SALS cases (Fig. 3a, b).The target genes in the spinal cord of amyotrophic lateral upregulation of SEL1L, another gene implicated in protein sclerosis cases degradation processes, indicates a likely activation of ATF6 (Yamamoto et al. 2007)(Fig.3b) while the activation of the The activation of the IRE1α-XBP1, together with the ATF6 IRE1α-XBP1 pathway was confirmed by the increase in the pathway, was investigated through the expression of the se- expression of DNAJB9 and DNAJC10 genes (Fig. 3c). lected 19 target genes (Table 1) in spinal cord samples derived Protein-protein interaction and gene ontology analysis re- from healthy individuals and sporadic cases of ALS (SALS). vealed that the upregulated genes in ALS cases were part of These genes were grouped in functional clusters as constitu- a close network functionally enriched in ERAD and - tive protein folding, chaperones and co-chaperones,anden- dependent protein catabolic processes terms (additional file: doplasmic reticulum-associated degradation (ERAD) based Fig. S3B-S3C). on their main function (Fig. 3). The expression of almost all Conversely, genes involved in the early stages of the con- the genes involved in ERAD such as HERPUD1, SEL1L, OS9 stitutive protein folding processes such as CANX, PDIA3 and Amyotrophic lateral sclerosis (ALS) and Alzheimer’s disease (AD) are characterised by differential...

comparable expression levels between control and SALS cases (Fig. 3e, additional file: Fig. S3A). In the chaperones and co-chaperones cluster, however, we found an elevated expression of HSPA5 gene, which encodes for a well- established marker of ER stress (BiP) and is regulated to the same extent by both TFs (Prell et al. 2012)(Fig.3e). Notably, we found a decrease in the expression of SIL1, a gene that is not regulated by either XBP1 or ATF6 and that encodes a nucleotide exchange factor for the ER chaperone HSPA5 (Fig. 3f). Activated IRE1α is also involved in the cleavage of meta- bolic mRNAs through a process known as IRE1α-dependent decay (RIDD) (Maurel et al. 2014; Tam et al. 2014;Hetzand Saxena 2017). RIDD is a pleiotropic mechanism affecting the expression of different sets of genes in different tissues, and the assessment of its role in neurodegenerative disease re- mains almost unknown. The major RIDD substrates identified in human and mouse are SCARA3 (Maurel et al. 2014;Tam et al. 2014), SPARC (Maurel et al. 2014; Tam et al. 2014)and BLOC1S1 (Maurel et al. 2014; Bright et al. 2015). In partic- ular, the latter has been hypothesised to be a possible universal RIDD target (Bright et al. 2015). Thus, we quantified the mRNAs of these genes in order to assess whether the RIDD activity of IRE1α is active in human post-mortem tissues. However, none of these genes showed a downregulation in the spinal cord of ALS (additional file: Fig. S2B). These re- sults indicate that ATF6 and IRE1α pathways activate a set of genes mainly involved in ERAD and protein degradation processes.

Fig. 2 Gene expression analysis of the spliced form of XBP1 (XBP1s) in the spinal cord and frontal and temporal cortex of amyotrophic lateral Characterisation of the expression of XBP1 target sclerosis (ALS), Alzheimer’s disease (AD) and frontotemporal lobar de- genes in the frontal and temporal cortex of cases mentia (FTLD) cases. a mRNA expression analysis of XBP1s in the of Alzheimer’sdisease spinal cord of healthy individuals (Ctrl, grey) and cases of amyotrophic lateral sclerosis (SALS, dark green). Scatter plots and bar plots are repre- sentations for the same samples. Ctrl, n = 9; ALS, n =17.b mRNA ex- After establishing that a significant increase occurred in pression analysis of XBP1s in the frontal and temporal cortex of healthy XBP1s levels (Fig. 2b), we investigated the expression of individuals (Ctrl, grey) and cases of Alzheimer’s disease (AD, orange). In the selected 19 genes in frontal and temporal regions of the the frontal cortex, Ctrl, n =19;AD,n = 14. In the temporal cortex, Ctrl, human brain derived from AD cases. Many of the genes up- n = 17; AD, n =14. c Gene expression analysis of the spliced form of XBP1 (XBP1s) in the frontal and temporal cortex of healthy individuals regulated in ALS that are involved in ERAD such as (Ctrl, grey) and patients with frontotemporal lobar dementia with (FTLD HERPUD1, SEL1L, PDIA4 and DNAJC10 also showed an C9+, red) and without (FTLD C9−, green) C9orf72 repeat expansions. In increased expression in one or both cortical regions of AD the frontal cortex, Ctrl, n =19;FTLDC9+,n =16;FTLDC9−, n =18.In cases (Fig. 4a, d). In contrast, the expression of OS9 and the temporal cortex, Ctrl, n =17;FTLDC9+,n =16;FTLDC9−, n =15. Means and SEMs were used to represent the data. The dots represent the DNAJB9, upregulated in SALS cases, was comparable be- samples analysed. According to the D’Agostino and Pearson normality tween disease and control cases (Fig. 4b, c). Another remark- test, all data are normally distributed. The unpaired t test or two-way able difference between AD and ALS cases was that the ex- ANOVA, followed by Tukey’s multiple comparison tests, was used. pression of protein disulphide isomerase (PDI) family mem- *P <0.05;**P < 0.01, ***P < 0.001. ALS, amyotrophic lateral sclerosis; AD, Alzheimer’s disease; FTLD, frontotemporal lobar dementia; Ctrl, bers involved in protein folding processes such as PDIA6 and control PDIA3, were selectively upregulated in AD (Fig. 4d) but not in ALS. Interestingly, a difference in gene expression also emerged between the cortical regions of AD cases. PDIA3 PDIA6 were not differentially expressed in ALS (Fig. 3d). Co- and CANX, two genes involved in the lectin folding cycle, chaperones involved in the folding processes such as were upregulated in the frontal cortex (Fig. 4d) while the ex- DNAJC3 and DNAJB11 (Shoulders et al. 2013) showed pression of HSPA5 and its co-chaperone DNAJC3 (Shoulders L. Montibeller, J. de Belleroche Amyotrophic lateral sclerosis (ALS) and Alzheimer’s disease (AD) are characterised by differential...

ƒFig. 3 Gene expression of XBP1 target genes tested in the spinal cord demonstrate that genes, which were upregulated in the spinal derived from sporadic cases of ALS (SALS). a–e Gene expression of a cord and temporal cortex of ALS and AD cases, show an in- representative group of XBP1 and ATF6 target genes in spinal cord post- mortem samples derived from healthy individuals (Ctrl, black) and cases crease also at the protein level. of amyotrophic lateral sclerosis (SALS, dark green). Genes were grouped by their main function in ERAD, endoplasmic reticulum-associated deg- radation (outlined in red), constitutive protein folding and chaperones Discussion (outlined in blue) and co-chaperones (outlined in green). f Gene expres- sion of a gene (SIL1) which is not regulated either by XBP1 or ATF6. Means and SEMs were used to represent the data. The dots represent The UPR is a network of signalling pathways that respond to ER individual samples and the numbers under the graphs represent the num- stress. Among the UPR signalling pathways, IRE1α and ATF6 ber of samples analysed. SALS, sporadic amyotrophic lateral sclerosis; predominantly play a pro-survival role through the Ctrl, control; No TFs, not regulated by transcription factors (XBP1 and/or ATF6). According to the D’Agostino and Pearson normality test, all data transactivation of the basic-leucine zipper (bZIP) transcription are normally distributed. The unpaired t test was used; *P < 0.05; factors, XBP1 and ATF6 (Yoshida et al. 2001; Adachi et al. **P < 0.01, ***P <0.001 2008). In this study, we found that IRE1α-XBP1 and ATF6 pathways activate different sets of genes in the spinal cord of et al. 2013), involved in protein folding processes, was in- SALS compared to frontal and temporal cortex of AD suggesting creased in the temporal cortex (Fig. 4e). ERLEC1 was also that different mechanisms of UPR activation are involved. differently expressed in temporal compared to the frontal cor- In the spinal cord of ALS cases, we found a substantial tex where it has been found to be downregulated (additional increase in the expression of genes associated with ERAD. file: Fig. 4SA). ERAD fulfils a crucial function through the clearance of The relevance of protein folding processes in AD was fur- misfolded proteins from the lumen of ER and, when impaired, ther confirmed by GO analysis of the upregulated genes in the results in accumulation of unfolded proteins (Redler and frontal and temporal cortex, which showed a functional en- Dokholyan 2012). The relevance of this process in ALS is richment in endoplasmic reticulum unfolded protein response further confirmed by the fact that several disease-causative and protein folding in endoplasmic reticulum stress terms (ad- mutations occur in genes involved in (e.g. VCP)oraffecting ditional file: Fig. S4B-C). Similarly to ALS, none of the RIDD (e.g. SOD1) these pathways (Boylan 2015). In fact, mutant targets showed a decrease in expression in AD cases, but SOD1 protein interacts specifically with Derlin-1, a compo- surprisingly, the expression of BLOC1S1 gene was found to nent of ERAD machinery, and triggers ER stress-induced ap- be increased in the frontal cortex (additional file: Fig. S2C). optosis in motor neurons through dysfunction of ERAD These results suggest a strong involvement of IRE1α-XBP1 (Nishitoh et al. 2008). We extended the investigation to other and ATF6 pathways, which induce the expression of genes ERAD components and we found that the expression of two involved in both protein degradation and protein folding such constitutive members of the ubiquin-degradation machinery, as PDI family members in AD. HERPUD1 and SEL1L, as well as two genes associated with ERAD, PDIA4 and OS9, was increased in human spinal cord of SALS cases. (Fig. 5a, b). In addition, we discovered that the Protein level analysis of representative genes expression of two co-chaperones, DNAJB9 and DNAJC10, upregulated or unchanged in the spinal cord was increased in disease cases. These co-chaperones belong and temporal cortex to the DNAJ family, which are proteins that can modulate the activity of HSPA5/BiP, the only chaperone present in Western blot (WB) analysis was carried out on a subset of pro- the ER (Kampinga and Bergink 2016). Notably, DNAJB9 and teins in the spinal cord and temporal cortex including genes that DNAJC10 are known to be associated with ERAD (Behnke were both upregulated or unchanged in order to substantiate et al. 2015) suggesting that their interaction with HSPA5/BiP changes found in mRNA. In the spinal cord, we analysed four promotes protein degradation processes and suppresses the proteins, three upregulated (HSPA5, DNAJC10, HERPUD1)and cell death induced by ER stress (Kurisu et al. 2003). This is one unchanged (PDIA6). Indeed, WB results were consistent further reinforced by the confirmation of the upregulation of with these findings. The expression of all three upregulated genes HSPA5/BiP, already shown by Atkin et al. (Atkin et al. 2008), showed a substantial increase also at the protein level (Fig. 6a, b) and DNAJC10 at the protein level in the spinal cord of SALS while, again consistent with qPCR results, PDIA6 showed no cases (Fig. 6). These results, in addition to the upregulation of change. The same proteins were investigated in the temporal HERPUD1 protein, indicate that the degradation machinery is cortex. Although fewer samples were analysed, WB results were upregulated at both mRNA and protein levels. Overloading of consistent with the qPCR findings also in this cortical region. The ERAD results in accumulation of unfolded proteins leading expression of the three upregulated genes (HSPA5, HERPUD1, the activation of the UPR. In this context, changes in the PDIA6)andDNAJC10, which was unchanged at the mRNA expression of SEL1L, a gene regulated exclusively by ATF6 level, showed a similar trend at the protein level. These results (Kaneko et al. 2007), and DNAJB9, a gene regulated L. Montibeller, J. de Belleroche Amyotrophic lateral sclerosis (ALS) and Alzheimer’s disease (AD) are characterised by differential...

ƒFig. 4 Gene expression analysis of XBP1 target genes tested in the A different profile of gene regulation was found in AD frontal and temporal cortex of the brain derived from Alzheimer’s whereby the upregulated genes were predominantly involved disease (AD) cases. a–e Gene expression of a representative group of XBP1 and ATF6 target genes in frontal and temporal cortex post- in multiple stages of protein homeostasis (Fig. 5a, b). As oc- mortem samples derived from healthy individuals (Ctrl, grey) and curs in ALS, we found an increased expression of ERAD Alzheimer’s disease cases (AD, orange). Genes were grouped by their genes such as PDIA4, HERPUD1, DNAJC10 and SEL1L in main function. f Gene expression of a gene (SIL1) which is not regulated frontal and/or in temporal cortical regions of AD cases. either by XBP1 or ATF6. Means and SEMs were used to represent the data. The dots represent the individual samples and the numbers under the Besides ERAD, we also found a profound activation of genes graphs represent the number of samples analysed. AD, Alzheimer’sdis- involved in protein folding processes. One example is given ease; Ctrl, control; No TFs, not regulated by transcription factors (XBP1 by the increase of HSPA5 expression in the temporal cortex and/or ATF6). According to the D’Agostino and Pearson normality test, which is associated with the upregulation of DNAJC3, a co- all data are normally distributed. The unpaired t test was used; *P <0.05; **P < 0.01, ***P <0.001 chaperone involved in early stage of protein folding (Shoulders et al. 2013;Behnkeetal.2015). In ALS, converse- ly, HSPA5 showed a co-upregulation with DNAJs involved in protein degradation, a disease difference worthy of further exclusively by XBP1 [3, 6, 49], suggest that these two UPR investigation. Another remarkable difference between gene pathways are co-activated in SALS cases. Interestingly, the expression profiles in these diseases is highlighted by the only downregulation observed in ALS cases was seen in the strong increase in the expression of several PDI family mem- expression of SIL1, a gene that is not regulated by either ATF6 bers in AD cases. Elevated expression of PDIA6, PDIA3, or XBP1. The SIL1 gene encodes for a nucleotide exchange PDIA4 and DNAJC10 (also known as PDIA19) emphasises factor associated with Hsp70 that facilitates substrate release the key role of PDI family members in cortical regions. PDI from BiP by stimulating the release of ADP (Behnke et al. family members play a crucial role in oxidative folding and 2015). Recently, de L’Etang et al. showed that SIL1 levels chaperone-mediated quality control of proteins by acting as a were selectively reduced in vulnerable motor neurons of general response when misfolded proteins start to accumulate SOD1-G93A mice (L’Etang et al. 2015). Thus, the reduced in the ER (Benham 2012). expression of SIL1, detected for the first time in human sam- Distinct gene expression profiles emerged also between the ples, could reflect the specific loss of motor neurons in the two cortical regions analysed. Genes encoding for proteins spinal cord of ALS cases. involved in the ER lectin cycle such as CANX and PDIA3

Fig. 5 Comparison between gene expression profiles in SALS and frontal and temporal cortex of AD cases. a Heat map of the gene expression levels of the 20 UPR target genes regulated by ATF6 and XBP1 and analysed in the lumbar spinal cord of SALS cases and the frontal and temporal cortex of AD cases. The gene expression was normalised against the controls and log2 FC was plotted. Each row in the heat maps corresponds to a tissue analysed; each column corresponds to a specific gene. The coloured bar indicates the range of intensity values for each gene in the heat maps. b Venn diagram for the upregulated genes found in SALS cases and AD cases. SALS, sporadic amyotrophic lateral sclerosis; AD temp Ctx, Alzheimer’sdisease temporal cortex; AD Fron Ctx, Alzheimer’s disease frontal cortex L. Montibeller, J. de Belleroche Amyotrophic lateral sclerosis (ALS) and Alzheimer’s disease (AD) are characterised by differential...

ƒFig. 6 Protein level analysis of representative genes upregulated or decay (RIDD) (Maurel et al. 2014; Tam et al. 2014; Hetz unchanged in the spinal cord and temporal cortex. a Western blot and Saxena 2017). Although it has been proposed that analyses are shown for HSPA5, DNAJC10, HREPUD1 and PDIA6 detected in the spinal cord of healthy individuals (Ctrl) and SALS cases RIDD can contribute to cell death during chronic conditions (SALS). Quantitative analyses of each protein were performed by densi- of ER stress (Maurel et al. 2014), the physiological role of this tometry of the relative bands in control and SALS cases. Values were process in neurodegenerative disease is still unclear. We found obtained using actin as a reference gene. Means and SEMs were used to that the expression of the major RIDD substrates was not represent the data. The numbers under the graphs represent the number of samples analysed. SALS, sporadic amyotrophic lateral sclerosis; Ctrl, altered in either ALS or AD. Although further investigations control. According to the D’Agostino and Pearson normality test, all data need to be performed in order to understand the role of RIDD are normally distributed. The unpaired t test was used; *P < 0.05. b in neurodegenerative diseases, we concluded that the RNase Representative western blots are shown for HSPA5, DNAJC10, activity of IRE1α does not affect the mRNA level of the HREPUD1 and PDIA6. c Western blot analyses are shown for HSPA5, DNAJC10 and PDIA6 detected in the temporal cortex of healthy induvial established RIDD targets in ALS and AD. Surprisingly, we (Ctrl) and AD cases (AD). Quantitative analyses of each protein were found that one of the putative RIDD target, BLOC1S1,was performed by densitometry of the relative bands in control and SALS profoundly upregulated in the frontal cortex of AD (additional cases. Values were obtained using actin as a reference gene. Means and file: Fig. S2B). This upregulation could indicate an activation SEMs were used to represent the data. The numbers under the graphs represent the number of samples analysed. AD, Alzheimer’sdisease;Ctrl, of UPR downstream pathways since BLOC1S1, in addition to control. The unpaired t test was used; *P <0.05.d Representative western being an RIDD substrate, is also involved in lysosomal bio- blots are shown for HSPA5, DNAJC10 and PDIA6 genesis (Pu et al. 2015) Several studies have demonstrated cooperation between ATF6 and IRE1α-XBP1 pathways (Yamamoto et al. 2007; were upregulated in the frontal cortex, while genes involved in Shoulders et al. 2013), and this is also confirmed by the partial chaperone and co-chaperone activity, such as HSPA5 and overlap of the biological processes associated with the target DNAJC3, were upregulated in the temporal cortex (Fig. 5a, genes regulated by these two (Fig. 1a, additional file: S1B). In b). Although both regions are characterised by changes in this context, there is evidence suggesting that the cooperative protein homeostasis during AD pathology (Hoozemans et al. effect of ATF6 and XBP1 is due to their ability to form het- 2005;Lindbergetal.2012;Mölleretal.2016), our results erodimers which have an increased affinity for the binding to indicate that either specific responses to ER stress are activat- specific cis-acting elements localised in the promoter regions ed in the frontal and temporal cortex or that the different re- of major UPR genes (Newman 2003; Yamamoto et al. 2007; gions reflect a different stage in disease progression. Shoulders et al. 2013). We showed that the majority of the Moreover, this evidence suggests that brain-region-specific genes that contain ERSE and ERSE-II response elements in factors must play a role in determining selective vulnerability their promoters are regulated by both transcription factors, such as the abundance of components of the protein folding while ACGT core and ERSE-26 elements appear to be regu- and quality control systems (Asuni et al. 2013;Jackson2014). lated specifically by XBP1 (Fig. 1d). Notably, these ERSE Notably, the majority of the changes in gene expression oc- elements are not a direct binding site of ATF4, the main tran- curred in the regions where the genes were expressed at lower scription factor activated by the PERK pathway (Ma et al. levels (Fig. 4). Moreover, HSPA5, DNAJC10 and PDIA6 2002). In addition, the UPRE consensus sequence appears to showed a trend to increase at the protein level confirming be associated with both TFs although a larger number of can- the gene expression results. In terms of magnitude of expres- didates need to be studied before a more robust conclusion can sion changes, the most robust upregulations were seen for be reached. Further investigations are also necessary to assess SEL1L and HERPUD1 in both AD and ALS indicating a the functional role of these consensus sequences in the regu- possible common mechanism of action. Finally, the dysregu- lation of each gene. lated expression of genes regulated by XBP1 and ATF6 indi- An interesting point of view was elaborated by Mori et al. cates that both ATF6 and IRE1α-XBP1 pathways are activat- who proposed a model whereby the ATF6-mediated response ed in both cortical regions of AD cases. The upregulation of is unidirectional (refolding only) while it is shifted by a XBP1- HERPUD1 and HSPA5, common genes regulated by all three mediated response to a bidirectional mode (refolding plus UPR pathways (Fig. 1a), could also reflect an activation of the degradation) depending on the nature of the stimulus PERK pathway. Although the PERK pathway has been found (Yoshida et al. 2003; Yamamoto et al. 2007). This model is to be involved in AD (Rozpedek et al. 2015;Yangetal.2016), also supported by the fact that ATF6 alone is still able to in this study, we focused on specific responses activated by induce the transcription of ER chaperones but not of ERAD IRE1α-XBP1 and ATF6 pathways which share several target components (Yamamoto et al. 2007). Thus, the substantial genes. In contrast the PERK pathway showed little overlap upregulation of genes found in ALS cases could indicate that with the other UPR pathways (additional file: Fig. S1B-D). the cooperation between ATF6α and XBP1 plays a crucial When activated, IRE1α is able to process XBP1 as well as role in the pathophysiological mechanism of this disease. In other mRNAs though a process called IRE1α-dependent AD, instead, the cooperative effect between these two L. Montibeller, J. de Belleroche pathways is less evident probably due to a strong contribution cell type- and condition-specific transcriptional regulatory networks. – of each of these pathways. Mol Cell 27:53 66. https://doi.org/10.1016/j.molcel.2007.06.011 Adachi Y, Yamamoto K, Okada T, Yoshida H, Harada A, Mori K (2008) ATF6 is a transcription factor specializing in the regulation of quality control proteins in the endoplasmic reticulum. Cell Struct Funct 33: 75–89. https://doi.org/10.1247/csf.07044 Conclusions Anagnostou G, Akbar MT, Paul P, Angelinetta C, Steiner TJ, de Belleroche J (2010) Vesicle associated membrane protein B (VAPB) is decreased in ALS spinal cord. Neurobiol Aging 31: In this study, we have compared the ER homeostatic pathways 969–985. https://doi.org/10.1016/j.neurobiolaging.2008.07.005 IRE1α-XBP1 and ATF6 in different neurodegenerative disor- Arai M, Kondoh N, Imazeki N, Hada A, Hatsuse K, Kimura F, Matsubara ders. Specifically, our results reveal that disease-specific pat- O, Mori K, Wakatsuki T, Yamamoto M (2006) Transformation- terns of IRE1α-XBP1 and ATF6 target genes are activated in associated gene regulation by ATF6α during hepatocarcinogenesis. – AD and ALS. The identification of these patterns provides a FEBS Lett 580:184 190. https://doi.org/10.1016/j.febslet.2005.11. 072 valuable source of information from which to develop new Asuni AA, Pankiewicz JE, Sadowski MJ (2013) Differential molecular selective therapeutic strategies in both ALS and AD. chaperone response associated with various mouse adapted scrapie Moreover, since UPR activation has been proposed to be ac- strains. Neurosci Lett 538:26–31. https://doi.org/10.1016/j.neulet. tivated in early stages of ALS and AD, the differential patterns 2013.01.027 Atkin JD, Farg MA, Walker AK, McLean C, Tomas D, Horne MK (2008) of gene activation in these disorders can be used as potential Endoplasmic reticulum stress and induction of the unfolded protein early biomarkers. response in human sporadic amyotrophic lateral sclerosis. Neurobiol Dis 30:400–407. https://doi.org/10.1016/j.nbd.2008.02.009 Acknowledgments We are grateful to the tissue donors and their relatives Behnke J, Feige MJ, Hendershot LM (2015) BiP and its nucleotide ex- involved in this research through the Imperial College ALS Tissue Bank change factors Grp170 and Sil1: mechanisms of action and biolog- and the Brains for Dementia Research Brain bank, King’s College ical functions. J Mol Biol 427:1589–1608. https://doi.org/10.1016/j. London. We thank other members of the lab Midhat Salman, Intan jmb.2015.02.011 Barul Akma Bakhtiar and Alex Morris for discussing gene expression Benham AM (2012) The protein disulfide isomerase family: key players experiments. in health and disease. Antioxid Redox Signal 16:781–789. https:// doi.org/10.1089/ars.2011.4439 Authors’ contributions L.M. and J.B. designed research, L.M. performed Boylan K (2015) Familial amyotrophic lateral sclerosis. Neurol Clin 33: research, J.B. contributed reagents/analytical tools, and L.M. and J.B. 807–830. https://doi.org/10.1016/j.ncl.2015.07.001 analysed data and wrote the manuscript. Bright MD, Itzhak DN, Wardell CP, Morgan GJ, Davies FE (2015) Cleavage of BLOC1S1 mRNA by IRE1 is sequence specific, tem- porally separate from XBP1 splicing, and dispensable for cell via- Funding Information This work was funded by a grant from EU H2020 bility under acute endoplasmic reticulum stress. Mol Cell Biol 35: MSCA ITN-675448 (TRAINERS). 2186–2202. https://doi.org/10.1128/MCB.00013-15 Casas-Tinto S, Zhang Y, Sanchez-Garcia J, Gomez-Velazquez M, Compliance with ethical standards Rincon-Limas DE, Fernandez-Funez P (2011) The ER stress factor XBP1s prevents amyloid- neurotoxicity. Hum Mol Genet 20:2144– The ALS component of this study was approved by the Riverside 2160. https://doi.org/10.1093/hmg/ddr100 Research Ethics Committee and was carried out according to their guide- Chen H-J, Anagnostou G, Chai A, Withers J, Morris A, Adhikaree J, lines. All tissue used in the Dementia component of this study was ob- Pennetta G, de Belleroche JS (2010) Characterization of the proper- tained from voluntary donors in compliance with the Mental Capacity Act ties of a novel mutation in VAPB in familial amyotrophic lateral (2005), and both Brain Banks have been approved by the National sclerosis. J Biol Chem 285:40266–40281. https://doi.org/10.1074/ Research Ethics Service. jbc.M110.161398 Cullinan SB, Zhang D, Hannink M, Arvisais E, Kaufman RJ, Diehl JA Conflict of interest The authors declare that they have no conflict of (2003) Nrf2 is a direct PERK substrate and effector of PERK- – interest. dependent cell survival. Mol Cell Biol 23:7198 7209 Dombroski BA, Nayak RR, Ewens KG, Ankener W, Cheung VG, Spielman RS (2010) Gene expression and genetic variation in re- Open Access This article is distributed under the terms of the Creative sponse to endoplasmic reticulum stress in human cells. Am J Hum Commons Attribution 4.0 International License (http:// Genet 86:719–729. https://doi.org/10.1016/j.ajhg.2010.03.017 creativecommons.org/licenses/by/4.0/), which permits unrestricted use, Dunham I, Kundaje A, Aldred SF et al (2012) An integrated encyclopedia distribution, and reproduction in any medium, provided you give of DNA elements in the . Nature 489:57–74. https:// appropriate credit to the original author(s) and the source, provide a link doi.org/10.1038/nature11247 to the Creative Commons license, and indicate if changes were made. Endres K, Reinhardt S (2013) ER-stress in Alzheimer’s disease: turning the scale? Am J Neurodegener Dis 2:247–265 Fels DR, Koumenis C (2006) The PERK/eIF2alpha/ATF4 module of the UPR in hypoxia resistance and tumor growth. Cancer Biol Ther 5: References 723–728 Filézac de L’etang A, Maharjan N, Braña MC et al (2015) Marinesco- Sjögren syndrome protein SIL1 regulates motor neuron subtype- Acosta-Alvear D, Zhou Y, Blais A, Tsikitis M, Lents NH, Arias C, selective ER stress in ALS. Nat Neurosci 18(2):227–238. https:// Lennon CJ, Kluger Y, Dynlacht BD (2007) XBP1 controls diverse doi.org/10.1038/nn.3903 Amyotrophic lateral sclerosis (ALS) and Alzheimer’s disease (AD) are characterised by differential...

Haze K, Yoshida H, Yanagi H, Yura T, Mori K (1999) Mammalian tran- Maurel M, Chevet E, Tavernier J, Gerlo S (2014) Getting RIDD of RNA: scription factor ATF6 is synthesized as a transmembrane protein and IRE1 in cell fate regulation. Trends Biochem Sci 39:245–254. activated by proteolysis in response to endoplasmic reticulum stress. https://doi.org/10.1016/j.tibs.2014.02.008 MolBiolCell10:3787–3799 Misiewicz M, Dery M-A, Foveau B et al (2013) Identification of a novel Hetz C, Saxena S (2017) ER stress and the unfolded protein response in endoplasmic reticulum stress response element regulated by XBP1. neurodegeneration. Nat Rev Neurol 13:477–491. https://doi.org/10. J Biol Chem 288:20378–20391. https://doi.org/10.1074/jbc.M113. 1038/nrneurol.2017.99 457242 Hetz C, Thielen P, Matus S, Nassif M, Court F, Kiffin R, Martinez G, Möller C, Hafkemeijer A, Pijnenburg YAL, Rombouts SARB, van der Cuervo AM, Brown RH, Glimcher LH (2009) XBP-1 deficiency in Grond J, Dopper E, van Swieten J, Versteeg A, Steenwijk MD, the nervous system protects against amyotrophic lateral sclerosis by Barkhof F, Scheltens P, Vrenken H, van der Flier WM (2016) increasing autophagy. Genes Dev 23:2294–2306. https://doi.org/10. Different patterns of cortical gray matter loss over time in behavioral 1101/gad.1830709 variant frontotemporal dementia and Alzheimer’sdisease. Hoozemans JJM, Veerhuis R, Van Haastert ES et al (2005) The unfolded Neurobiol Aging 38:21–31. https://doi.org/10.1016/j. protein response is activated in Alzheimer’sdisease.Acta neurobiolaging.2015.10.020 Neuropathol 110:165–172. https://doi.org/10.1007/s00401-005- Nardo G, Pozzi S, Pignataro M, Lauranzano E, Spano G, Garbelli S, 1038-0 Mantovani S, Marinou K, Papetti L, Monteforte M, Torri V, Paris Hoozemans JJM, van Haastert ES, Nijholt DAT, Rozemuller AJM, L, Bazzoni G, Lunetta C, Corbo M, Mora G, Bendotti C, Bonetto V Eikelenboom P, Scheper W (2009) The unfolded protein response (2011) Amyotrophic lateral sclerosis multiprotein biomarkers in pe- is activated in pretangle neurons in Alzheimer’s disease hippocam- ripheral blood mononuclear cells. PLoS One 6:e25545. https://doi. pus. Am J Pathol 174:1241–1251. https://doi.org/10.2353/ajpath. org/10.1371/journal.pone.0025545 2009.080814 Newman JRS (2003) Comprehensive identification of human bZIP inter- Howarth DL, Lindtner C, Vacaru AM, Sachidanandam R, Tsedensodnom actions with coiled-coil arrays. Science 300:2097–2101. https://doi. O, Vasilkova T, Buettner C, Sadler KC (2014) Activating transcrip- org/10.1126/science.1084648 tion factor 6 is necessary and sufficient for alcoholic fatty liver Nishitoh H, Kadowaki H, Nagai A, Maruyama T, Yokota T, Fukutomi H, disease in zebrafish. PLoS Genet 10:e1004335. https://doi.org/10. Noguchi T, Matsuzawa A, Takeda K, Ichijo H (2008) ALS-linked 1371/journal.pgen.1004335 mutant SOD1 induces ER stress- and ASK1-dependent motor neu- Ilieva EV, Ayala V, Jove M, Dalfo E, Cacabelos D, Povedano M, ron death by targeting Derlin-1. Genes Dev 22:1451–1464. https:// Bellmunt MJ, Ferrer I, Pamplona R, Portero-Otin M (2007) doi.org/10.1101/gad.1640108 Oxidative and endoplasmic reticulum stress interplay in sporadic Paul P, Murphy T, Oseni Z, Sivalokanathan S, de Belleroche JS (2014) amyotrophic lateral sclerosis. Brain 130:3111–3123. https://doi. Pathogenic effects of amyotrophic lateral sclerosis-linked mutation org/10.1093/brain/awm190 in D-amino acid oxidase are mediated by D-serine. Neurobiol Aging Jackson WS (2014) Selective vulnerability to neurodegenerative disease: 35:876–885. https://doi.org/10.1016/j.neurobiolaging.2013.09.005 the curious case of prion protein. Dis Model Mech 7:21–29. https:// Pincus D, Chevalier MW, Aragón T, van Anken E, Vidal SE, el-Samad H, doi.org/10.1242/dmm.012146 Walter P (2010) BiP binding to the ER-stress sensor Ire1 tunes the Kampinga HH, Bergink S (2016) Heat shock proteins as potential targets homeostatic behavior of the unfolded protein response. PLoS Biol 8: for protective strategies in neurodegeneration. Lancet Neurol 15: e1000415. https://doi.org/10.1371/journal.pbio.1000415 748–759. https://doi.org/10.1016/S1474-4422(16)00099-5 Prell T, Lautenschläger J, Witte OW, Carri MT, Grosskreutz J (2012) The Kaneko M, Yasui S, Niinuma Y, Arai K, Omura T, Okuma Y, Nomura Y unfolded protein response in models of human mutant G93A amyo- (2007) A different pathway in the endoplasmic reticulum stress- trophic lateral sclerosis. Eur J Neurosci 35:652–660. https://doi.org/ induced expression of human HRD1 and SEL1 genes. FEBS Lett 10.1111/j.1460-9568.2012.08008.x 581:5355–5360. https://doi.org/10.1016/j.febslet.2007.10.033 Kanemoto S, Kondo S, Ogata M, Murakami T, Urano F, Imaizumi K Pu J, Schindler C, Jia R, Jarnik M, Backlund P, Bonifacino JS (2015) (2005) XBP1 activates the transcription of its target genes via an BORC, a multisubunit complex that regulates lysosome positioning. – ACGT core sequence under ER stress. Biochem Biophys Res Dev Cell 33:176 188. https://doi.org/10.1016/j.devcel.2015.02.011 Commun 331:1146–1153. https://doi.org/10.1016/j.bbrc.2005.04. Redler RL, Dokholyan NV (2012) The complex molecular biology of 039 amyotrophic lateral sclerosis (ALS). Prog Mol Biol Transl Sci 107: – Katayama T, Imaizumi K, Honda A, Yoneda T, Kudo T, Takeda M, Mori 215 262. https://doi.org/10.1016/B978-0-12-385883-2.00002-3 K, Rozmahel R, Fraser P, George-Hyslop PS, Tohyama M (2001) Rozpedek W, Markiewicz L, Diehl JA, Pytel D, Majsterek I (2015) Disturbed activation of endoplasmic reticulum stress transducers by Unfolded protein response and PERK kinase as a new therapeutic familial Alzheimer’s disease-linked presenilin-1 mutations. J Biol target in the pathogenesis of Alzheimer’s disease. Curr Med Chem Chem 276:43446–43454. https://doi.org/10.1074/jbc.M104096200 22:3169–3184 Kurisu J, Honma A, Miyajima H, Kondo S, Okumura M, Imaizumi K Shoulders MD, Ryno LM, Genereux JC, Moresco JJ, Tu PG, Wu C, Yates (2003) MDG1/ERdj4, an ER-resident DnaJ family member, sup- JR III, Su AI, Kelly JW, Wiseman RL (2013) Stress-independent presses cell death induced by ER stress. Genes Cells 8:189–202 activation of XBP1s and/or ATF6 reveals three functionally diverse Lee JH, Won SM, Suh J, Son SJ, Moon GJ, Park UJ, Gwag BJ (2010) ER proteostasis environments. Cell Rep 3:1279–1292. https://doi. Induction of the unfolded protein response and cell death pathway in org/10.1016/j.celrep.2013.03.024 Alzheimer’s disease, but not in aged Tg2576 mice. Exp Mol Med Song Y, Sretavan D, Salegio EA, Berg J, Huang X, Cheng T, Xiong X, 42:386–394. https://doi.org/10.3858/emm.2010.42.5.040 Meltzer S, Han C, Nguyen TT, Bresnahan JC, Beattie MS, Jan LY, Lindberg O, Westman E, Karlsson S, Östberg P, Svensson LA, Simmons Jan YN (2015) Regulation of axon regeneration by the RNA repair A, Wahlund LO (2012) Is the subcallosal medial prefrontal cortex a and splicing pathway. Nat Neurosci 18:817–825. https://doi.org/10. common site of atrophy in Alzheimer’s disease and frontotemporal 1038/nn.4019 lobar degeneration? Front Aging Neurosci 4:32. https://doi.org/10. SriburiR,BommiasamyH,BuldakGL,RobbinsGR,FrankM, 3389/fnagi.2012.00032 Jackowski S, Brewer JW (2007) Coordinate regulation of phospho- Ma Y, Brewer JW, Diehl JA, Hendershot LM (2002) Two distinct stress lipid biosynthesis and secretory pathway gene expression in XBP- signaling pathways converge upon the CHOP promoter during the 1(S)-induced endoplasmic reticulum biogenesis. J Biol Chem 282: mammalian unfolded protein response. J Mol Biol 318:1351–1365 7024–7034. https://doi.org/10.1074/jbc.M609490200 L. Montibeller, J. de Belleroche

Tam AB, Koong AC, Niwa M (2014) Ire1 has distinct catalytic mecha- and XBP1. Dev Cell 13:365–376. https://doi.org/10.1016/j.devcel. nisms for XBP1/HAC1 splicing and RIDD. Cell Rep 9:850–858. 2007.07.018 https://doi.org/10.1016/j.celrep.2014.09.016 Yang W, Zhou X, Zimmermann HR, Cavener DR, Klann E, Ma T (2016) Travers KJ, Patil CK, Wodicka L, Lockhart DJ, Weissman JS, Walter P Repression of the eIF2α kinase PERK alleviates mGluR-LTD im- (2000) Functional and genomic analyses reveal an essential coordi- pairments in a mouse model of Alzheimer’s disease. Neurobiol nation between the unfolded protein response and ER-associated Aging 41:19–24. https://doi.org/10.1016/j.neurobiolaging.2016.02. degradation. Cell 101:249–258 005 Xiang C, Wang Y, Zhang H, Han F (2017) The role of endoplasmic Ye J, Rawson RB, Komuro R, Chen X, Davé UP, Prywes R, Brown MS, reticulum stress in neurodegenerative disease. Apoptosis 22:1–26. Goldstein JL (2000) ER stress induces cleavage of membrane- https://doi.org/10.1007/s10495-016-1296-4 bound ATF6 by the same proteases that process SREBPs. Mol Yamamoto K, Yoshida H, Kokame K, Kaufman RJ, Mori K (2004) Cell 6:1355–1364 Differential contributions of ATF6 and XBP1 to the activation of Yoshida H, Matsui T, Yamamoto A, Okada T, Mori K (2001) XBP1 endoplasmic reticulum stress-responsive cis-acting elements ERSE, mRNA is induced by ATF6 and spliced by IRE1 in response to UPRE and ERSE-II. J Biochem 136:343–350. https://doi.org/10. ER stress to produce a highly active transcription factor. Cell 107: 1093/jb/mvh122 881–891. https://doi.org/10.1016/S0092-8674(01)00611-0 Yamamoto K, Sato T, Matsui T, Sato M, Okada T, Yoshida H, Harada A, Yoshida H, Matsui T, Hosokawa N, Kaufman RJ, Nagata K, Mori K Mori K (2007) Transcriptional induction of mammalian ER quality (2003) A time-dependent phase shift in the mammalian unfolded control proteins is mediated by single or combined action of ATF6α protein response. Dev Cell 4:265–271